nast.cpp

   1 /*
   2  *  nast.cpp
   3  *
   4  *
   5  *  Created by Pat Schloss on 12/17/08.
   6  *  Copyright 2008 Patrick D. Schloss. All rights reserved.
   7  *
   8  *      This is my implementation of the NAST (nearest alignment space termination) algorithm as described in:
   9  *
  10  *      DeSantis TZ, Hugenholtz P, Keller K, Brodie EL, Larsen N, Piceno YM, Phan R, & Anderson GL.  2006.  NAST: a multiple
  11  *              sequence alignment server for comparative analysis of 16S rRNA genes.  Nucleic Acids Research.  34:W394-9.
  12  *
  13  *      To construct an object one needs to provide a method of getting a pairwise alignment (alignment) and the template
  14  *      and candidate sequence that are to be aligned to each other.
  15  *
  16  */
  17
  18 #include "sequence.hpp"
  19 #include "alignment.hpp"
  20 #include "nast.hpp"
  21
  22 /**************************************************************************************************/
  23
  24 Nast::Nast(Alignment* method, Sequence* cand, Sequence* temp) : alignment(method), candidateSeq(cand), templateSeq(temp) {
  25         try {
  26                 maxInsertLength = 0;
  27                 pairwiseAlignSeqs();    //      This is part A in Fig. 2 of DeSantis et al.
  28                 regapSequences();               //      This is parts B-F in Fig. 2 of DeSantis et al.
  29
  30         }
  31         catch(exception& e) {
  32                 errorOut(e, "Nast", "Nast");
  33                 exit(1);
  34         }
  35
  36 }
  37
  38 /**************************************************************************************************/
  39
  40 void Nast::pairwiseAlignSeqs(){ //      Here we call one of the pairwise alignment methods to align our unaligned candidate
  41                                                                 //      and template sequences
  42         try {
  43
  44                 alignment->align(candidateSeq->getUnaligned(), templateSeq->getUnaligned());
  45
  46                 string candAln = alignment->getSeqAAln();
  47                 string tempAln = alignment->getSeqBAln();
  48
  49                 if(candAln == ""){
  50
  51                         candidateSeq->setPairwise("");
  52                         templateSeq->setPairwise(templateSeq->getUnaligned());
  53
  54                 }
  55                 else{
  56                         if(tempAln[0] == '-'){
  57                                 int pairwiseAlignmentLength = tempAln.length(); //      we need to make sure that the candidate sequence alignment
  58                                 for(int i=0;i<pairwiseAlignmentLength;i++){             //      starts where the template sequence alignment starts, if it
  59                                         if(isalpha(tempAln[i])){                                        //      starts early, we nuke the beginning of the candidate
  60                                                 candAln = candAln.substr(i);                    //      sequence
  61                                                 tempAln = tempAln.substr(i);
  62                                                 break;
  63                                         }
  64                                 }
  65                         }
  66                         int pairwiseAlignmentLength = tempAln.length();
  67                         if(tempAln[pairwiseAlignmentLength-1] == '-'){          //      we need to make sure that the candidate sequence alignment
  68                                 for(int i=pairwiseAlignmentLength-1; i>=0; i--){//      ends where the template sequence alignment ends, if it runs
  69                                         if(isalpha(tempAln[i])){                                        //      long, we nuke the end of the candidate sequence
  70                                                 candAln = candAln.substr(0,i+1);
  71                                                 tempAln = tempAln.substr(0,i+1);
  72                                                 break;
  73                                         }
  74                                 }
  75                         }
  76
  77                 }
  78
  79                 candidateSeq->setPairwise(candAln);                                     //      set the pairwise sequences in the Sequence objects for
  80                 templateSeq->setPairwise(tempAln);                                      //      the candidate and template sequences
  81
  82         }
  83         catch(exception& e) {
  84                 errorOut(e, "Nast", "pairwiseAlignSeqs");
  85                 exit(1);
  86         }
  87 }
  88
  89 /**************************************************************************************************/
  90
  91 void Nast::removeExtraGaps(string& candAln, string tempAln, string newTemplateAlign){
  92
  93 //      here we do steps C-F of Fig. 2 from DeSantis et al.
  94         try {
  95
  96                 int longAlignmentLength = newTemplateAlign.length();
  97
  98                 for(int i=0; i<longAlignmentLength; i++){                               //      use the long alignment as the standard
  99                         int rightIndex, rightRoom, leftIndex, leftRoom;
 100
 101                         //      Part C of Fig. 2 from DeSantis et al.
 102                         if((isalpha(newTemplateAlign[i]) != isalpha(tempAln[i]))){      //if there is a discrepancy between the regapped
 103
 104                                 rightRoom = 0; leftRoom = 0;
 105
 106                                 //      Part D of Fig. 2 from DeSantis et al.           //      template sequence and the official template sequence
 107                                 for(leftIndex=i-1;leftIndex>0;leftIndex--){     //      then we've got problems...
 108                                         if(!isalpha(candAln[leftIndex])){
 109                                                 leftRoom = 1;   //count how far it is to the nearest gap on the LEFT side of the anomaly
 110                                                 while(leftIndex-leftRoom>=0 && !isalpha(candAln[leftIndex-leftRoom]))   {       leftRoom++;             }
 111                                                 break;
 112                                         }
 113                                 }
 114
 115                                 for(rightIndex=i+1;rightIndex<longAlignmentLength-1;rightIndex++){
 116                                         if(!isalpha(candAln[rightIndex])){
 117                                                 rightRoom = 1;  //count how far it is to the nearest gap on the RIGHT side of the anomaly
 118                                                 while(rightIndex+rightRoom<longAlignmentLength && !isalpha(candAln[rightIndex+rightRoom]))      {       rightRoom++;    }
 119                                                 break;
 120                                         }
 121                                 }
 122
 123                                 int insertLength = 0;                                                   //      figure out how long the anomaly is
 124                                 while(!isalpha(newTemplateAlign[i + insertLength]))     {       insertLength++; }
 125                                 if(insertLength > maxInsertLength){     maxInsertLength = insertLength; }
 126
 127                                 if((leftRoom + rightRoom) >= insertLength){
 128
 129                                         //      Parts D & E from Fig. 2 of DeSantis et al.
 130                                         if((i-leftIndex) <= (rightIndex-i)){            //      the left gap is closer - > move stuff left there's
 131
 132                                                 if(leftRoom >= insertLength){                   //      enough room to the left to move
 133
 134                                                         string leftTemplateString = newTemplateAlign.substr(0,i);
 135                                                         string rightTemplateString = newTemplateAlign.substr(i+insertLength);
 136                                                         newTemplateAlign = leftTemplateString + rightTemplateString;
 137                                                         longAlignmentLength = newTemplateAlign.length();
 138
 139                                                         string leftCandidateString = candAln.substr(0,leftIndex-insertLength+1);
 140                                                         string rightCandidateString = candAln.substr(leftIndex+1);
 141                                                         candAln = leftCandidateString + rightCandidateString;
 142
 143                                                 }
 144                                                 else{                                                                   //      not enough room to the left, have to steal some space to
 145                                                         string leftTemplateString = newTemplateAlign.substr(0,i);       //      the right
 146                                                         string rightTemplateString = newTemplateAlign.substr(i+insertLength);
 147                                                         newTemplateAlign = leftTemplateString + rightTemplateString;
 148                                                         longAlignmentLength = newTemplateAlign.length();
 149
 150                                                         string leftCandidateString = candAln.substr(0,leftIndex-leftRoom+1);
 151                                                         string insertString = candAln.substr(leftIndex+1,rightIndex-leftIndex-1);
 152                                                         string rightCandidateString = candAln.substr(rightIndex+(insertLength-leftRoom));
 153                                                         candAln = leftCandidateString + insertString + rightCandidateString;
 154
 155                                                 }
 156                                         }
 157                                         else{                                                                           //      the right gap is closer - > move stuff right there's
 158                                                 if(rightRoom >= insertLength){                  //      enough room to the right to move
 159
 160                                                         string leftTemplateString = newTemplateAlign.substr(0,i);
 161                                                         string rightTemplateString = newTemplateAlign.substr(i+insertLength);
 162                                                         newTemplateAlign = leftTemplateString + rightTemplateString;
 163                                                         longAlignmentLength = newTemplateAlign.length();
 164
 165                                                         string leftCandidateString = candAln.substr(0,rightIndex);
 166                                                         string rightCandidateString = candAln.substr(rightIndex+insertLength);
 167                                                         candAln = leftCandidateString + rightCandidateString;
 168
 169                                                 }
 170                                                 else{                                                                   //      not enough room to the right, have to steal some
 171                                                         //      space to the left lets move left and then right...
 172                                                         string leftTemplateString = newTemplateAlign.substr(0,i);
 173                                                         string rightTemplateString = newTemplateAlign.substr(i+insertLength);
 174                                                         newTemplateAlign = leftTemplateString + rightTemplateString;
 175                                                         longAlignmentLength = newTemplateAlign.length();
 176
 177                                                         string leftCandidateString = candAln.substr(0,leftIndex-(insertLength-rightRoom)+1);
 178                                                         string insertString = candAln.substr(leftIndex+1,rightIndex-leftIndex-1);
 179                                                         string rightCandidateString = candAln.substr(rightIndex+rightRoom);
 180                                                         candAln = leftCandidateString + insertString + rightCandidateString;
 181
 182                                                 }
 183                                         }
 184                                 }
 185                                 else{
 186                                                                                                         //      there could be a case where there isn't enough room in
 187                                         string leftTemplateString = newTemplateAlign.substr(0,i);                       //      either direction to move stuff
 188                                         string rightTemplateString = newTemplateAlign.substr(i+leftRoom+rightRoom);
 189                                         newTemplateAlign = leftTemplateString + rightTemplateString;
 190                                         longAlignmentLength = newTemplateAlign.length();
 191
 192                                         string leftCandidateString = candAln.substr(0,leftIndex-leftRoom+1);
 193                                         string insertString = candAln.substr(leftIndex+1,rightIndex-leftIndex-1);
 194                                         string rightCandidateString = candAln.substr(rightIndex+rightRoom);
 195                                         candAln = leftCandidateString + insertString + rightCandidateString;
 196
 197                                 }
 198
 199                                 i -= insertLength;
 200                         }
 201                 }
 202         }
 203         catch(exception& e) {
 204                 errorOut(e, "Nast", "removeExtraGaps");
 205                 exit(1);
 206         }
 207 }
 208
 209 /**************************************************************************************************/
 210
 211 void Nast::regapSequences(){    //This is essentially part B in Fig 2. of DeSantis et al.
 212         try {
 213
 214                 string candPair = candidateSeq->getPairwise();
 215                 string candAln = "";
 216
 217                 string tempPair = templateSeq->getPairwise();
 218                 string tempAln = templateSeq->getAligned();             //      we use the template aligned sequence as our guide
 219
 220                 int pairwiseLength = candPair.length();
 221                 int fullAlignLength = tempAln.length();
 222
 223                 if(candPair == ""){
 224                         for(int i=0;i<fullAlignLength;i++)      {       candAln += '.';         }
 225                         candidateSeq->setAligned(candAln);
 226                         return;
 227                 }
 228
 229                 int fullAlignIndex = 0;
 230                 int pairwiseAlignIndex = 0;
 231                 string newTemplateAlign = "";                                   //      this is going to be messy so we want a temporary template
 232                 //      alignment string
 233                 while(tempAln[fullAlignIndex] == '.' || tempAln[fullAlignIndex]  == '-'){
 234                         candAln += '.';                                                         //      add the initial '-' and '.' to the candidate and template
 235                         newTemplateAlign += tempAln[fullAlignIndex];//  pairwise sequences
 236                         fullAlignIndex++;
 237                 }
 238
 239                 string lastLoop = "";
 240
 241                 while(pairwiseAlignIndex<pairwiseLength){
 242                         if(isalpha(tempPair[pairwiseAlignIndex]) && isalpha(tempAln[fullAlignIndex])
 243                            && isalpha(candPair[pairwiseAlignIndex])){
 244                                 //  the template and candidate pairwise and template aligned have characters
 245                                 //      need to add character onto the candidatSeq.aligned sequence
 246
 247                                 candAln += candPair[pairwiseAlignIndex];
 248                                 newTemplateAlign += tempPair[pairwiseAlignIndex];//
 249
 250                                 pairwiseAlignIndex++;
 251                                 fullAlignIndex++;
 252                         }
 253                         else if(isalpha(tempPair[pairwiseAlignIndex]) && !isalpha(tempAln[fullAlignIndex])
 254                                         && isalpha(candPair[pairwiseAlignIndex])){
 255                                 //      the template pairwise and candidate pairwise are characters and the template aligned is a gap
 256                                 //      need to insert gaps into the candidateSeq.aligned sequence
 257
 258                                 candAln += '-';
 259                                 newTemplateAlign += '-';//
 260                                 fullAlignIndex++;
 261                         }
 262                         else if(!isalpha(tempPair[pairwiseAlignIndex]) && isalpha(tempAln[fullAlignIndex])
 263                                         && isalpha(candPair[pairwiseAlignIndex])){
 264                                 //  the template pairwise is a gap and the template aligned and pairwise sequences have characters
 265                                 //      this is the alpha scenario.  add character to the candidateSeq.aligned sequence without progressing
 266                                 //      further through the tempAln sequence.
 267
 268                                 candAln += candPair[pairwiseAlignIndex];
 269                                 newTemplateAlign += '-';//
 270                                 pairwiseAlignIndex++;
 271                         }
 272                         else if(isalpha(tempPair[pairwiseAlignIndex]) && isalpha(tempAln[fullAlignIndex])
 273                                         && !isalpha(candPair[pairwiseAlignIndex])){
 274                                 //  the template pairwise and full alignment are characters and the candidate sequence has a gap
 275                                 //      should not be a big deal, just add the gap position to the candidateSeq.aligned sequence;
 276
 277                                 candAln += candPair[pairwiseAlignIndex];
 278                                 newTemplateAlign += tempAln[fullAlignIndex];//
 279                                 fullAlignIndex++;
 280                                 pairwiseAlignIndex++;
 281                         }
 282                         else if(!isalpha(tempPair[pairwiseAlignIndex]) && !isalpha(tempAln[fullAlignIndex])
 283                                         && isalpha(candPair[pairwiseAlignIndex])){
 284                                 //      the template pairwise and aligned are gaps while the candidate pairwise has a character
 285                                 //      this would be an insertion, go ahead and add the character->seems to be the opposite of the alpha scenario
 286
 287                                 candAln += candPair[pairwiseAlignIndex];
 288                                 newTemplateAlign += tempAln[fullAlignIndex];//
 289                                 pairwiseAlignIndex++;
 290                                 fullAlignIndex++;
 291                         }
 292                         else if(isalpha(tempPair[pairwiseAlignIndex]) && !isalpha(tempAln[fullAlignIndex])
 293                                         && !isalpha(candPair[pairwiseAlignIndex])){
 294                                 //      template pairwise has a character, but its full aligned sequence and candidate sequence have gaps
 295                                 //      this would happen like we need to add a gap.  basically the opposite of the alpha situation
 296
 297                                 newTemplateAlign += tempAln[fullAlignIndex];//
 298                                 candAln += "-";
 299                                 fullAlignIndex++;
 300                         }
 301                         else if(!isalpha(tempPair[pairwiseAlignIndex]) && isalpha(tempAln[fullAlignIndex])
 302                                         && !isalpha(candPair[pairwiseAlignIndex])){
 303                                 //      template and candidate pairwise are gaps and the template aligned is not a gap this should not be possible
 304                                 //      would skip the gaps and not progress through full alignment sequence
 305                                 //      not tested yet
 306
 307                                 mothurOut("We're into D " + toString(fullAlignIndex) + " " +  toString(pairwiseAlignIndex)); mothurOutEndLine();
 308                                 pairwiseAlignIndex++;
 309                         }
 310                         else{
 311                                 //      everything has a gap - not possible
 312                                 //      not tested yet
 313
 314                                 mothurOut("We're into F " +  toString(fullAlignIndex) + " " +  toString(pairwiseAlignIndex)); mothurOutEndLine();
 315                                 pairwiseAlignIndex++;
 316                                 fullAlignIndex++;
 317                         }
 318                 }
 319
 320                 for(int i=fullAlignIndex;i<fullAlignLength;i++){
 321                         candAln += '.';
 322                         newTemplateAlign += tempAln[i];//
 323                 }
 324
 325                 int start, end;
 326                 for(int i=0;i<candAln.length();i++){
 327                         if(candAln[i] == 'Z' || !isalnum(candAln[i]))   {       candAln[i] = '.';       }       //      if we padded the alignemnt from
 328                         else{                   start = i;                      break;          }                                                       //      blast with Z's, change them to
 329                 }                                                                                                                                                               //      '.' characters
 330
 331                 for(int i=candAln.length()-1;i>=0;i--){                                                                                 //      ditto.
 332                         if(candAln[i] == 'Z' || !isalnum(candAln[i]))   {       candAln[i] = '.';       }
 333                         else{                   end = i;                        break;          }
 334                 }
 335
 336                 for(int i=start;i<=end;i++){                                    //      go through the candidate alignment sequence and make sure that
 337                         candAln[i] = toupper(candAln[i]);                       //      everything is upper case
 338                 }
 339
 340
 341                 if(candAln.length() != tempAln.length()){               //      if the regapped candidate sequence is longer than the official
 342                         removeExtraGaps(candAln, tempAln, newTemplateAlign);//  template alignment then we need to do steps C-F in Fig.
 343                 }                                                                                               //      2 of Desantis et al.
 344
 345
 346                 candidateSeq->setAligned(candAln);
 347         }
 348         catch(exception& e) {
 349                 errorOut(e, "Nast", "regapSequences");
 350                 exit(1);
 351         }
 352 }
 353
 354 /**************************************************************************************************/
 355
 356 float Nast::getSimilarityScore(){
 357         try {
 358
 359                 string cand = candidateSeq->getAligned();
 360                 string temp = templateSeq->getAligned();
 361                 int alignmentLength = temp.length();
 362                 int mismatch = 0;
 363                 int denominator = 0;
 364
 365                 for(int i=0;i<alignmentLength;i++){
 366                         if(cand[i] == '-' && temp[i] == '-'){
 367
 368                         }
 369                         else if(cand[i] != '.' && temp[i] != '.'){
 370                                 denominator++;
 371
 372                                 if(cand[i] != temp[i]){
 373                                         mismatch++;
 374                                 }
 375                         }
 376                 }
 377                 float similarity = 100 * (1. - mismatch / (float)denominator);
 378                 if(denominator == 0){   similarity = 0.0000;    }
 379
 380                 return similarity;
 381
 382         }
 383         catch(exception& e) {
 384                 errorOut(e, "Nast", "getSimilarityScore");
 385                 exit(1);
 386         }
 387 }
 388
 389 /**************************************************************************************************/
 390
 391 int Nast::getMaxInsertLength(){
 392
 393         return maxInsertLength;
 394
 395 }
 396
 397 /**************************************************************************************************/